Learning and predicting time series by neural networks
Authors:
Preprint
Abstract
Artificial neural networks which are trained on a time series are supposed to achieve two abilities: firstly to predict the series many time steps ahead and secondly to learn the rule which has produced the series. It is shown that prediction and learning are not necessarily related to each other. Chaotic sequences can be learned but not predicted while quasiperiodic sequences can be well predicted but not learned.Keywords
All Related Versions
- Version 1, 2002-02-28, ArXiv
- Published version: Physical Review E, 65 (5), 050903.
This publication has 5 references indexed in Scilit:
- Robust chaos generation by a perceptronEurophysics Letters, 2000
- Time series generation by multilayer networksPhysical Review E, 1998
- Limit cycles of a perceptronJournal of Physics A: General Physics, 1998
- Analytical Study of Time Series Generation by Feed-Forward NetworksPhysical Review Letters, 1995
- Generation and Prediction of Time Series by a Neural NetworkPhysical Review Letters, 1995